lvm warning message

[curt larson_1]

when I try to extend a lvol to create a mirror i get this warning message:
Warning: logical volume number xxxx found in physical volume not found in /dev/vg01.

I've attached a few old discs to this workstation to do some lvm testing. I noticed one already had a boot partition. So, I thought I'd import it and see if there is already an OS installed. the import went fine. I tried import a second disc and vgimport wasn't able to find any lvm information. So, I did a pvcreate -B on this disc. and then a vgextend to add it to vg01 So, I was going to mirror this drive to the one I had imported. everything seems to work fine, except I get the above warning message whenever i do a lvextend to create a mirror or a lvreduce to remove the mirror.

any clues to the cause or a way to stop getting the message.

thanks

[melvyn burnard]

Hmm, you may need to check your existing device files and make sure you do not have a duplicated lvol minor number.

[curt larson_1]

checked the lvol device files. no duplicates. lvols named lvol1-8 with minor numbers 0x01000[1-8] with matching character devices files rlvol1-8 with matching minor numbers 0x01000[1-8]
no extras, none missing.

pvdisplay shows 8 lvols and vgdisplay shows 8 lvols

[Patrick Wallek]

Are you saying that your minor #'s on the group files are:

0x010001
0x010002, etc?

If so that's incorrect. The minor #'s should be:

0x010000
0x020000
0x030000, etc......

[S.K. Chan]

That's the error you would get for none existing device files or similar device files. The minor numbers you indicated are correct (for vg01) ..
0x010000 = group
0x010001 = lvol1 and rlovl1
0x010002 = lvol2 and rlvol2
The lvol that you're having the problem extending, try to recreate its device files .. for instance if it's lvol2..
# mknod /dev/vg01/rlvol2 c 64 0x010002
# mknod /dev/vg01/lvol2 b 64 0x010002
What about the rest of the lvols ? Do you get the same error when you extend it ? Another thing I noticed ..why pvcreate with the -B option, you meant -f I hope.

[curt larson_1]

I just didn't want to type in a bunch of lvols, but to make it more clear, I'v attached an ll of my /dev/vg01 directory.

I've tried removing lvol8 and rlvol8 and recreating the nodes. No change. Still getting same warning message when doing a lvreduce to remove a mirror or a lvextend to add a mirror copy.

I did get the warning message when extending every lvol to add a mirrored copy. But, I've been doing my testing (making changes) just to lvol8 so far.

adding the second disk. vgimport reported no lvm structures on the disk. I did a pvcreate -B (no -f option was given). pvcreate finished sucessfully, with no messages of their being previous lvm structues on the disk being destroyed.

[curt larson_1]

looks like the lvlnboot info is incorrect. updated the b, r, s, and d info with correct information and the warning has gone away

[Michael Steele_2]

Regarding:

...So, I was going to mirror this drive to the one I had imported...

Could you provide the syntax of the vgexport and vgimport please?

Also, could you now create another map file and check it against your old map file as well as your /etc/lvmtab file? With specific attention to the disks and VGID?

vgexport -m mapfile -p /dev/vg01

strings /etc/lvmtab | more

[curt larson_1]

Could you provide the syntax of the vgexport and vgimport please?

I don't know the prior history of the discs. Maybe they were exported, maybe not. No, I didn't have a mapfile for importing the disc.

vgimport /dev/vg01 /dev/dsk/c1t6d0

vgexport -m mapfile -p /dev/vg01

strings /etc/lvmtab | more

the attached file has this information.

[Michael Steele_2]

Careful with vgexport for it will remove the volume group from /etc/lvmtab and the system without the "...-p..." arguement.

1)vgchange -a n /dev/vg##
2)vgexport -p -v -s -m mapfile /dev/vg##
3)vgexport -v -s -m mapfile /dev/vg## (* NOTE: This will update /etc/lvmtab and remove the volume group *)

For vgimport use:

a)mkdir /dev/vg##
b)mknod /dev/vg##/group -c 64 0x0#0000 (* NOTE: 0x0#0000 must be unique *)
c)vgimport -s -p -m mapfile /dev/vg## (*NOTE: This will preview and display responding disks *)
d1)vgimport -s -m mapfile /dev/vg## (* Create *)
d2)Alternative if d1) fails:
vgimport /dev/vg## /dev/dsk/c#t#d0 (* NOTE: As you have done *)

I'm looking for a disconnect with the VGID and the disks associated to the VG and /etc/lvmtab. They should all match up.